Initial SQL-only 1C adapter baseline

scripts/download_missing_hf_models.py

@@ -0,0 +1,61 @@
+from __future__ import annotations
+
+import argparse
+import os
+import sys
+
+from huggingface_hub import snapshot_download
+
+from plan_hf_downloads import remote_plan
+from common import iter_model_card_paths, read_yaml_mapping
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description="Download missing Hugging Face model weights from model cards.")
+    parser.add_argument("--model", action="append", dest="models", help="Model card id to download. Can be repeated.")
+    parser.add_argument("--include-ready", action="store_true")
+    parser.add_argument("--dry-run", action="store_true")
+    args = parser.parse_args()
+
+    selected = set(args.models or [])
+    token = os.environ.get("HF_TOKEN") or None
+    plans = []
+    for card_path in iter_model_card_paths():
+        card = read_yaml_mapping(card_path)
+        card_id = str(card.get("id") or "")
+        if selected and card_id not in selected:
+            continue
+        plan = remote_plan(card_path, include_ready=args.include_ready)
+        if plan and plan["missing_bytes"]:
+            plans.append(plan)
+
+    if not plans:
+        print("No missing downloads.")
+        return 0
+
+    for plan in plans:
+        missing_files = [item["name"] for item in plan["files"] if not item["complete"]]
+        if not missing_files:
+            continue
+        print("")
+        print(f"== {plan['id']} ==")
+        print(f"Repository: {plan['repo_id']}")
+        print(f"Destination: {plan['storage_path']}")
+        print(f"Missing: {plan['missing_gb']} GB")
+        for name in missing_files:
+            print(f"- {name}")
+        if args.dry_run:
+            continue
+        snapshot_download(
+            repo_id=plan["repo_id"],
+            local_dir=plan["storage_path"],
+            token=token,
+            allow_patterns=missing_files,
+            ignore_patterns=["*.msgpack", "*.h5", "*.ot", "*.tflite", "*.onnx", "*.pb", "*.ckpt"],
+        )
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())